Strategy

  • Start small — begin with test jobs to determine actual resource needs
  • Match resources to requirements — request only the cores and memory your job will use
  • Consider wait times — smaller requests often queue faster
  • Use shared queues — for jobs that don't need a full node
  • Verify compatibility — ensure software works with the target CPU/GPU architecture
Note: Memory only needs to be specified for shared queues; otherwise all available node memory is requested by default.

Legacy queues (login1/login2) — 28-core Intel Haswell

AVX2 support, 128 GB memory per node.

QueueDurationMax nodesSharedBest for
debug-28core1 hr8NoTesting/debugging
short-28core1–4 hr12NoSmall/medium jobs
medium-28core4–12 hr24NoMedium parallel jobs
long-28core8–48 hr8NoLong simulations
extended-28core8 hr – 7 days2NoVery long simulations
large-28core4–8 hr80NoLarge parallel jobs

Legacy GPU queues

QueueGPUGPU memoryDurationBest for
gpu4× K8024 GB1–8 hrBasic GPU workloads
gpu-long4× K8024 GB8–48 hrLong GPU jobs
p1002× P10016 GB1–24 hrScientific computing
v1002× V10032 GB1–24 hrML/AI

Modern queues (milan1/milan2) — 40-core Intel Skylake

AVX512 support, 192 GB memory per node.

QueueDurationMax nodesSharedBest for
debug-40core1 hr8NoTesting/debugging
short-40core1–4 hr8NoStandard jobs
short-40core-shared1–4 hr4YesSmaller jobs
medium-40core4–12 hr16NoMedium jobs
long-40core8–48 hr6NoLong jobs
long-40core-shared8–24 hr3YesShared long jobs
extended-40core8 hr – 7 days2NoVery long jobs
extended-40core-shared8 hr – 3.5 days1YesShared extended jobs
large-40core4–8 hr50NoLarge parallel jobs

96-core AMD EPYC Milan

AVX2 support, 256 GB memory per node.

QueueDurationMax nodesSharedBest for
short-96core1–4 hr8NoParallel jobs
short-96core-shared1–4 hr4YesModerate shared jobs
medium-96core4–12 hr16NoParameter sweeps
long-96core8–48 hr6NoLong parallel jobs
long-96core-shared8–24 hr3YesShared long jobs
extended-96core8 hr – 7 days2NoVery long jobs
extended-96core-shared8 hr – 3.5 days1YesShared extended jobs
large-96core4–8 hr38NoLarge parallel jobs

High-bandwidth memory (HBM) nodes

AMX/AVX512 support, 384 GB per node (256 GB DDR5 + 128 GB HBM). Use HBM nodes for workloads limited by memory bandwidth.

QueueDurationMax nodesSpecial featuresBest for
hbm-short-96core1–4 hr8High-bandwidth memoryMemory-intensive jobs
hbm-medium-96core4–12 hr16Enhanced memory performanceLarge datasets
hbm-long-96core8–48 hr62–4× memory speedMemory-bound simulations
hbm-extended-96core8 hr – 7 days2Maximum durationLong memory jobs
hbm-large-96core4–8 hr38Large-scale HBM computingMassive memory-bound jobs
hbm-1tb-long-96core8–48 hr11 TB memory + 128 GB HBM cacheExtremely large datasets

NVIDIA A100 GPU nodes

4× A100 80 GB GPUs on 64-core Intel Ice Lake nodes (256 GB memory). Use GPU queues only for compatible applications — verify CUDA/software requirements before submission.

QueueDurationMax nodesSharedBest for
a1001–8 hr2YesGPU workloads, AI/ML
a100-long8–48 hr1YesLong GPU jobs
a100-large1–8 hr4YesLarge GPU jobs
Applies to SeaWulf